Variables dépendantes

Les variables dépendantes ont été regroupées en trois catégories. Le comportement navigationnel a été évalué par : 1) la vitesse moyenne vers l’avant du point milieu du fauteuil roulant, 2) le dégagement minimal (plus petite distance en centimètres entre le corps ou le fauteuil roulant et la tige verticale ou l’obstacle), et 3) le premier et deuxième (si présent) point de déviation de la trajectoire du point milieu du fauteuil roulant, en relation avec la condition LD, calculé comme étant le temps avant de franchir la tige verticale ou l’obstacle. De plus, la présence ou non d’un contre-mouvement a été quantifié. La coordination temporelle a été évaluée par : 4) la coordination temporelle des déviations angulaires des yeux, de la tête, du tronc et du fauteuil roulant vers la nouvelle direction, calculé en temps avant de franchir la tige vertical ou l’obstacle, et 5) le temps relatif mesuré par la différence entre la coordination temporelle entre les yeux et la tête, la tête et le tronc, et le tronc et le fauteuil roulant. Basé sur les travaux de Hollands et collègues (Hollands et al., 2002), le comportement visuel a été divisé en quatre grands comportements, soient : a) les fixations dirigées sur la tige verticale ou l’obstacle, la cible, le plancher ou autres éléments de l’environnement, b) les saccades, c) les mouvements des yeux avec la tête, incluant les fixations de déplacements et les mouvements de la tête, puis d) les données manquantes. Une description plus détaillée de chaque comportement visuel se trouve au tableau 2. Chaque comportement visuel était rapporté en pourcentage du comportement visuel total et les valeurs moyennes des essais selon la condition ont été calculées. Toutes ces variables dépendantes ont été analysées durant la phase d’approche des trois conditions, soit à partir de la position de départ (début du corridor) des sujets jusqu’au point de croisement de la tige verticale ou de l’obstacle, à l’exception de la variable vitesse moyenne du fauteuil roulant où la zone d’accélération à l’intérieur du corridor était exclus des calculs.

Tableau 2. Description de la variable comportement visuel

Catégories et comportements Description Durée

a. Fixations

1.1 Tige verticale ou obstacle

Stabilisation du regard sur un objet ou un endroit dans l’environnement. ≥ 3 échantillons (99.9 msec) 1.2 Cible 1.3 Plancher 1.4 Autres éléments de l’environnement b. Saccades Mouvement(s) rapide(s) entre deux endroits ou objets dans

l’environnement.

__

c. Mouvements des yeux avec la tête

3.1 Fixation de déplacement

Stabilisation du regard à une distance constante sur le plancher en avant du sujet et dont le point de regard se déplace avec le sujet. ≥ 3 échantillons (99.9 msec) 3.2 Mouvement de la tête

Les yeux et la tête tournent

simultanément. __

d. Données manquantes

4.1 Clignement des

yeux Absence du curseur rouge __

4.2 Comportement indéterminé

Comportement qui ne concorde dans aucune catégorie décrite ci-haut

__

2.4.3 Procédures statistiques

Tout d’abord, des analyses descriptives (moyennes et écart-types) ont été réalisées pour toutes les variables dépendantes ainsi que pour les variables sociodémographiques et les résultats aux questionnaires WST et WheelCon. Des tests-T ont été utilisés afin de comparer les caractéristiques sociodémographiques de nos deux échantillons. Pour l’objectif principal, un ANOVA à mesures répétées a été utilisé pour analyser la vitesse [2 groupes x 3 conditions], le dégagement minimal et le point de déviation initial de la trajectoire [2 groupes x 2 conditions pour ces deux variables], la coordination temporelle [2

21

groupes x 2 conditions x 4 segments] et le comportement visuel [2 groupes x 3 conditions x 6 comportements visuels]. Un test-T a été utilisé pour une comparaison simple pour la variable du temps relatif de la coordination du corps. Pour la présence du contre- mouvement, un test à 2 échantillons pour l’égalité des proportions a été utilisé pour comparer les groupes. Trois usagers en fauteuil roulant manuel et un en fauteuil roulant motorisé avaient des mouvements oculaires trop angulés pour le système Mobile Eye XG, résultant en un nombre important de données manquantes. Ainsi, nous avons exclus ces sujets des analyses descriptives et statistiques pour les trois variables dépendantes impliquant des données visuelles, soient la coordination temporelle, le temps relatif et le comportement visuel. De plus, lors de l’acquisition des données, nous avons réalisé qu’il n’y avait pas de mouvements des yeux vers la nouvelle direction lors du contournement d’obstacle pour tous les participants. Ainsi, des ANOVAS à mesures répétées ont été réalisées séparément pour les deux conditions, soit pour le CD [2 groupes x 4 segments] et pour le CO [2 groupes x 3 segments].

Pour l’objectif secondaire, qui vise à évaluer l’expérience sur le contrôle visuo- locomoteur sous-jacent à la navigation en fauteuil roulant, des corrélations de Pearson ont été utilisées pour analyser l’effet des années d’expérience, des habiletés perçus en fauteuil roulant (WST) et de la confiance d’utiliser le fauteuil roulant (WheelCon) sur les différentes variables dépendantes. Le logiciel SPSS (version 23.0) a été utilisé pour réaliser les tests statistiques, avec un niveau de signification établi à 0,05.

Chapitre 3 : Article 1

3.1 Résumé

Introduction. Les fauteuils roulants (FR) manuels ou motorisés sont fréquemment utilisés chez les personnes ayant une lésion de la moelle épinière pour assister la mobilité. Pourtant, le contrôle visuo-locomoteur sous-jacent à la navigation en FR chez les usagers expérimentés n’est pas bien compris à ce jour et est nécessaire afin de mieux informer les cliniciens pour soutenir l’entraînement des habiletés en FR ainsi que le développement des technologies d’assistance à la navigation.

Objectif. L’objectif de cette étude était de comparer le contrôle visuo-locomoteur en FR manuel et motorisé chez des sujets ayant une lésion de la moelle épinière lors de changement de direction et de contournement d’obstacle.

Méthodes. Les participants ayant une lésion de la moelle épinière utilisant un FR manuel (n=12, 38,5±10,7 ans) ou motorisé (n=10, 47,8±8,6 ans) devaient manœuvrer leur FR à une vitesse naturelle en ligne droite, lors de changement de direction de 45° vers la droite et lors de contournement d’obstacle vers la droite. La vitesse, le dégagement minimal du corps, le point de déviation, la coordination temporelle du corps, le temps relatif de la coordination temporelle entre les segments et le comportement visuel ont été analysés. Résultats. Aucun effet majeur de groupe n’a été trouvé pour la vitesse, le dégagement minimal et le point de déviation. Lors du changement de direction, la tête initiait toujours la réorientation du corps et du FR tandis que lors du contournement d’obstacle, une stratégie «en bloc» a été utilisée par les deux groupes. Lors de la navigation en ligne droite, les participants fixaient principalement la cible finale. Durant le changement de direction et le contournement d’obstacle, les participants fixaient davantage la trajectoire future et les obstacles pour les deux modes de locomotion.

Conclusion. Les usagers en FR manuel et motorisé adoptent des stratégies navigationnelles et une coordination temporelle visuo-locomotrice similaire lors de changement de direction et de contournement d’obstacle. Le comportement visuel spécifique dépend des exigences environnementales. Les études futures devraient regarder des tâches locomotrices plus complexes (ex. : avec des obstacles en mouvement).

23

3.2 Article

Visuo-locomotor control in persons with spinal cord injury in a manual or power wheelchair for direction changes and obstacles circumvention

Caroline Charette, PT1,2_{, François Routhier, PEng, PhD}1,2_{, Bradford J. McFadyen, PhD}1,2

1. Centre for Interdisciplinary Research in Rehabilitation and Social Integration (CIRRIS), Institut de réadaptation en déficience physique de Québec, CIUSSS-CN, Quebec, Canada 2. Faculty of Medicine, Departement of Rehabilitation, Université Laval, Quebec, Canada

Corresponding author:

Bradford J. McFadyen, PhD, CIRRIS, 525 Boul. Wilfrid-Hamel, Quebec, Canada, G1M 2S8, Email: brad.mcfadyen@fmed.ulaval.ca, Tel.: 1-418-529-9141 x6584

3.2.1 Abstract

Background. Manual (MWC) or power (PWC) wheelchairs are frequently used in individuals with a spinal cord injury (SCI) to assist mobility. Yet, visuo-locomotor control underlying WC navigation in experienced users is not well understood and necessary to better inform WC skills training as well as the development of assistive navigation technologies.

Objective. The objective was to compare the visuo-locomotor control for MWC and PWC in subjects with SCI while changing direction and circumventing an obstacle.

Methods. Participants with SCI using a MWC (n=12, 38.5±10.7y) or a PWC (n=10, 47.8±8.6y) were asked to maneuver at self-selected speed straight ahead, while changing direction 45° to the right, and while circumventing an obstacle to the right. Speed, minimal clearance, point of deviation, temporal body and WC coordination, relative timing of segment rotations and visual behavior were analysed.

Results. There was no main effect of group for speed, clearance and point of deviation. During direction changes, the head always lead body and wheelchair reorientation while an ‘en bloc’ strategy was used for circumventing obstacles for both groups. In straight- ahead locomotion, participants predominantly fixed their gaze on the end target. During direction changes and obstacle circumvention, participants fixated more on the future path and the obstacle for both WC modes.

Conclusions. MWC and PWC users adopt similar navigational strategies and visuo- locomotor coordination while changing direction and circumventing obstacles. Specific gaze behavior depends on environmental demands. Future studies should look at more complex WC locomotor tasks (e.g., with moving obstacles).

25 3.2.2 Introduction

Wheelchairs users are estimated at 3.6 million among community-dwelling individuals in the USA (Brault 2012) and over 285,000 in Canada (Smith et al. 2016). These numbers are likely to increase in the next decades in relation to the aging of the population (Canada 2015; Colby 2015). Individuals with spinal cord injury (SCI), make up a large proportion of WC users. In the USA, it is estimated that there are over 276,000 individuals living with SCI (NSCISC 2015), while in Canada, more than 85,000 people have a SCI (Noonan et al. 2012). The American National Spinal Cord Injury Statistical Center has identified that approximately 58% of individuals with SCI use a WC during the first year following their injury, increasing up to 80% after thirty years post-injury. From this sample, the manual wheelchair (MWC) has been identified as the most common type of WC used, but power wheelchair (PWC) use increases to as high as 42% with time post- injury (NSCISC 2015).

In everyday life, wheelchair users must be able to adapt their locomotion to environmental constraints in order to navigate efficiently and safely, but such steering control is achieved differently for the MWC (bi-manual propulsion) versus the PWC (joystick manipulation). To date, research on wheeled mobility has mainly focused on evaluating and training wheelchair skills (Kirby et al. 2015; Routhier et al. 2012) or on the biomechanics of propulsion (Blouin et al. 2015; Gagnon et al. 2014; Lalumiere et al. 2014).

While studies on PWC navigation have mostly been interested in smart PWCs to assist navigation, a recent study reported that experienced PWC users still encounter difficulties within the context of community mobility for tasks such as avoiding static and dynamic (e.g., people) obstacles (Torkia et al. 2015). The authors suggested, among other things, that better WC training by clinicians could benefit patients in better anticipating problematic situations while navigating in their community including better planning of obstacle

avoidance (Torkia et al. 2015). However, to our knowledge, WC skills training and the development of assistive navigation technologies do not appear to be based on the natural visuo-locomotor behavior of experienced WC users. Due to differences related to chair dimensions, body movements and modes of steering, we expect that the navigation of the manual and power WCs would be different.

Adapting locomotion to the environment relies on the visual system to provide spatio-temporal information. A good perception of the environment contributes to the planning of locomotor adjustments (anticipatory strategy) allowing more effective

interaction with the environment (Higuchi 2013). It was demonstrated that experienced WC users had better perception of spatial requirements for whether a doorway was passable with their WC, even despite somatosensory deficits in individuals with tetraplegia (Higuchi et al. 2009b).

Visuo-locomotor control during direction changes has been more widely studied for biped locomotion demonstrating a robust pre-planned motor pattern that occurs within two steps prior to a direction change. Specifically, the head always leads the rotation to the new direction, followed by the trunk, the pelvis and then the feet (Bernardin et al. 2012; Grasso et al. 1996; Grasso et al. 1998; Hicheur et al. 2005; Hollands et al. 2002; Hollands et al. 2001; Imai et al. 2001; Patla et al. 1999). This top-down body reorientation strategy has been observed for different angles of direction change, including turning a corner triggered by visual and auditory cues (Grasso et al. 1998; Imai et al. 2001; Pradeep Ambati et al. 2013). However, few studies have looked at the visuo-locomotor strategies while maneuvering a wheelchair for such tasks. In a recent study, walking and MWC navigation were compared in healthy participants while changing direction (Charette et al. 2015). It was demonstrated that these novice WC users also adopted a top-down strategy with rotation of the head leading the change of direction. The novice MWC users also combined a counter-movement of the WC prior to the time of path deviation towards the

27

target (Charette et al. 2015) similar to what has been observed for turning an automobile (Mars, 2008). For those novice MWC users, the counter-movement was likened to that observed for vehicular direction changes and thought to contribute to greater clearance of the intermediate poll that was used to indicate point of direction change. It is not known, however, if such vehicular-like counter-movement is also used by experienced MWC and PWC users.

For its part, locomotor control during biped obstacle circumvention has been less studied. In healthy individuals, all body segments appear to turn together while

circumventing an obstacle (Paquette and Vallis 2010; Vallis and McFadyen 2003). The change in the walking path appears to be initiated as soon as the participants started walking, but there was a second, more pronounced, path deviation made at approximately 1.5 m from the obstacle as a final locomotor adjustment (Gerin-Lajoie et al. 2005). To our knowledge, locomotor adjustments during obstacle circumvention have not been studied while maneuvering a wheelchair.

For gaze behavior during adaptive locomotion, either while walking (Higuchi 2013; Hollands et al. 2002) or maneuvering a wheelchair (Charette et al. 2015; Higuchi et al. 2009a), it has been demonstrated that the majority of fixations are directed towards the future travel path or on an object of interest according to the task. Most studies have reported saccadic eye movements towards a target at the end-point of the new travel path that preceded head rotation, which was hypothesized to provide information regarding the timing and the magnitude of body segment rotations needed to perform the direction change (Higuchi 2013; Hollands et al. 2002). Higuchi et al (2009a) also demonstrated that the fixation patterns were dependent on the form of locomotion that individuals were familiar with. They found in healthy persons that fixations were directed more frequently towards door edges while maneuvering a WC than while walking. To our knowledge, no

study has looked at visual behavior of experienced WC users, such as individuals with a SCI, during navigation tasks.

Despite the projected increase in WC users and advances in WC technology for assisted navigation, understanding visuo-locomotor control for WC locomotion in relation to manual and powered modes of propulsion by populations using such transport within daily environmental contexts is lacking. In addition to furthering our understanding of general locomotor navigation, such knowledge also has the potential to assist clinicians in both training and prescribing WCs as well as engineers in designing sophisticated

navigational aids for WCs. The purpose of the present study was to compare the visuo- locomotor control for MWC and PWC navigation, during direction change and obstacle circumvention in individuals with a spinal cord injury, a population for which WC

locomotion is often essential and, therefore, could serve as a gold standard for WC use. Based on the dominant top-down visuo-locomotor coordination observed during biped locomotion that appears to be transferred to the WC by novice users, we hypothesized that experienced manual and power wheelchair users will show similar body and WC temporal coordination within tasks. However, due to the differences in relation to chair dimensions, body movements and modes of steering, we also hypothesis that different anticipatory path control (temporal points of deviation and counter-movements), as well as different gaze behavior will be used between MWC and PWC modes within tasks.

3.2.3 Methods

Participants

A convenience sample of 12 MWC and 10 PWC users with SCI were recruited. Ethics approval was obtained from the Institut de réadaptation en déficience physique de Québec and all participants provided written informed consent prior to the experiment. The inclusion criteria were to: (1) be between the age of 18 and 55 years; (2) have sustained a

29

SCI (paraplegia for MWC; tetraplegia or paraplegia for PWC) since the age of 15 years; (3) use a WC as a primary means of mobility for at least one year (> 5 hours/day); and (4) propel their WC bimanually (MWC) or with the use of a joystick (PWC). Participants who had self-reported shoulder pain (visual analogue scale > 3/10) while maneuvering a WC, other neurological conditions or a score below 20/20 on the Snellen visual acuity test were excluded.

Experimental procedures

Participants first completed a socio-demographic questionnaire (age, sex, self- reported SCI neurological level, type and year of injury, years of experience in WC). Tire pressures were verified and adjusted to the manufacturer recommended pressure if needed. Then, participants were asked to maneuver their WC at a natural self-selected speed for three environmental conditions (see Fig. 1; 5 trials/condition): (1) straight-ahead along an 8.75m path (SA condition), (2) direction change (DC condition) of 45° to the right, indicated by a black vertical pole (VP; 1.86m height, 3.5cm diameter) and (3)

circumvention on the right side of a cylindrical obstacle (OC condition; 1.8m height, 43 cm diameter). For all conditions, the participants were required to maneuver their WC to a target placed at the end of each path (see arrows in Fig. 1). The center points of the VP and the cylindrical obstacle were located 3.6 meters from the starting position. A corridor (0.92m wide, 1.85m long) made of two black wooden planks was placed at the starting point to encourage an initial straight-line acceleration zone. The vertical pole for DC (Fig 1b) was aligned with the right side of the corridor and the center of cylindrical obstacle for OC (Fig 1c) was aligned with the center of the initial pathway.

Data collection and outcome measures

Kinematic data of body and WC movements were collected using three

triads of non-collinear infrared makers placed on the head, sternal notch, forearms and front of the WC frame. Eye movements were collected using a commercial eye tracker (Mobile Eye XG, Applied Sciences Laboratories, 30 Hz), which was synchronized to the motion analysis system.

Visuo-locomotor control was described under three categories. Navigational

behavior was defined by the average forward speed of the midpoint of the WC, the

minimal clearance (smallest distance between the body or the WC and the VP or obstacle), the point of first and second (if any) deviation of the midpoint of the WC in relation to mean SA trajectory calculated as the time (in negative values) before crossing the VP/obstacle (that was assigned as time zero). In addition, the presence or not of a counter-movement for the DC condition was quantified. Temporal coordination was assessed by the onset of angular deviations of the eye, head and trunk or WC towards the new travel path relative to the time of crossing the VP/obstacle and the relative timing of movement between adjacent segments measured as the difference in temporal

coordination between eye and head, head and trunk, and trunk and WC. Finally, visual

behavior was characterized by video coding, using a frame-by-frame analysis with the

PhysMo Video motion analysis software. In accordance with Hollands et al. (2002), visual behavior was categorized as: a) fixations on a location or object within the scene (≥3 frames), specifically on the target, the VP or obstacle, the floor, or any other environmental features (OEF); b) saccades, which are rapid eye movements between two locations; (c) eye movements following head including travel fixations where gaze stabilizes at a constant distance and moves with the participants (≥3 frames), as well as head

movements where the eye and head turn together; d) missing data, including blinks and those that could not be categorized as noted above. Each category of gaze behavior was reported as a percentage of total visual behavior and mean values of trials for each